Today, Rahul Goma Phulore (aka @missingfaktor) quoted Aristotle [1],

'Not to have one meaning is to have no meaning, and if words have no meaning, our reasoning with one another, and indeed with ourselves, has been annihilated.'

and, as often, I disagreed.

(No, I didn't check the quote. I want to talk about its content, not its historical correctness. So I don't care.)

So I read the quote as: I should have one opinion, and not many competing ones.

Variations of this thesis can be found often. I found, e.g., a funny version in a book about the freedom of will, where Peter Bieri [2] said, to believe something contradictory (and it neads more than one proposition to get a contradiction, thats the connection to the quote) means to believe nothing at all. This made me laugh: If I'd use the words this way, there may be people who never thought or believed anything. Arrogant? Maybe, but as we know since Gödel [3], I cannot be sure that I did either.

To argue here, I have to specify what I refer to when using the word "opinion".
I will use "having an opinion" as synonym to "having a model about a situation, with the possibility to use it in a decision process". There are no doubt many more meanings or facets of meaning to the word, but to explain my position that it is good to have more than one opinion (on one single topic, I haste to add), it suffices to show it for a partial meaning.

The crucial distinction I want to introduce into the discussion is the on between thinking and acting.

Of course, you cannot act (in the very physical sense) one one way A, and at the same time the other way B that contradicts A. Why physical sense? Because I can very well, for example, do something that is good for my health in one aspect, and bad for it in another. So, things like evaluations, meanings, and all that belong to a realm where contradiction is not only possible, but even to be expected. Lets leave that aside.

At least you cannot move your left arm from left to right and from right to left at the same moment in the same coordinate system.

So there is a sense, or level of description, where we can say, you can do exactly one action, and no additional (contradicting) one. Thats about action.

But if you plan to act, you do not act in this very sense. You ponder what to do. It is here where a multitude of models comes in. If the above quote is to be understood that I should have only one theory / one model, about how to act, then it is harmful. The crucial point is that we usually cannot be sure what model is the right one. Even worse, the concept of "the right one" may be ill defined. What we have instead is some model, and some experience about them fitting reality.

Let me mention two metaphors. I go with the radical constructivists here, btw. [4]

1. A model can be seen as a tool. A hammer isn't true or false - it's useful, applicable, and all this, but not true.

2. The second metaphor is about the distinction between a key and a picklock. Of course, if you know the lock, the distinction is clear: The key is the right one, the one that got delivered by the producer of the lock. The picklock is anything that is produced by somebody else. So if you are not sure, the producer will be. We say the key matches, the picklock just fits.

Well, but let's apply this metaphor to models and theories: If you dont introduce the assumtion that there be a constructor of the object in scrutiny, the key part of the metaphor fails to apply, there is no theory that matches the object - there are only fitting ones, only picklocks, and we can try to compare their fitting in certain situations.

(And yes, the coincidence with darwinian fitness is intended.)

Well, a metaphor doesnt prove anything. But it may lead you to a new way of seeing something.

In constructivist epistemology the concept of truth is mainly reduced to internal consistency inside of a theory.

Interestingly, that again fits the situation we have , e.g., in physics. We did not abandon Newton mechanics after Einstein, we know where to use it still; and where to use one of the relativity theories or quantum mechanics, or any combination.

But physics may be an example that actually leads us down the garden path, simply because it is mathematicalised in a way that we even can hope for a single theory. In daily situations we most often dont know what to think, dont know what model to base our decisions on, but have a buch of competing hypotheses. And thats the situation I want to point at. If we dont know which model to use, we should NOT AT ALL abandon all models except the most probable. Because this probability changes with time (we call this "learning"; ehm, well, yes, there's more to it). Having only one model is tantamount to not having any doubt in it. Thats exactly the contrary to what science is about. [5]

In the very moment you use these models to act, you have to choose one of the models to lead you [6]. And I think the one with the highes probability will do. But it is a bias often seen in humans that they assume, after using a model, they have to commit to its correctness. No, you dont. Yesterdays best decision is still yesterdays best decision; even after you learned today, and updated your state of knowledge, and got a new best decision. Thats todays. You could not know that yesterday. [7]

The quality of a decision process is orthogonal to the knowledge base it is applied upon. [8]

Humans are notoriously bad at keeping more than two models about a certain topic in their mind. Some seem to have difficulties to handle more than one. So this may be a daring task.

Nevermind, give it a try. (And just in case you build a nonhuman intelligence, please dont make it so stupid to use only one model.)

So the very moment of action is the moment where one of the models is chosen to act upon. Nevertheless, the full knowledge representation is kept, and developed. A model is only abandoned when it is surely refuted. And a refutation means: Proven to be useless in future. Thats rather hard to proof, by the way. (Think at Newtons mechanics. Yes, its wrong, but nevertheless, its used every day.)

Ah, one hint. Is what I wrote here true or false? Thats a question I often get, when talking about constructivist epistemology. The answer is: No. Its usable, and I found it helpful.

Another hint. There are people who deny the true/false distinction even in sutuations where it is completely clear what everybody is talking about. This is useless and annoying, to say the least. (Here I would use another metaphor: Even if the the border called "dawn" is rather fuzzy, the distinction between day and night is helpful. [9]) [10]

[1] http://www.twitlonger.com/show/ap93s1

[2] Peter Bieri, "Das Handwerk der Freiheit: Über die Entdeckung des eigenen Willens", e.g. p22 (German only, sorry.)

[3] Kurt Gödel: Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I. in: Monatshefte für Mathematik und Physik. Akademische Verlagsgesellschaft, Leipzig 38.1931, pp. 173-198. (An English translation by Martin Hirzel can be found here: http://www.research.ibm.com/people/h/hirzel/papers/canon00-goedel.pdf )

[4] Wikipedia, http://en.wikipedia.org/wiki/Radical_constructivism (Not one of the brilliant articles, sad enough)

[5] E.T. Jaynes, "Probability Theory: The Logic of Science". (Suggested reading)

[6] Even this can be disputed: You may, e.g., use model A, to lead the action, while contradicting as least as possible against models B and C, or something.

[7] If you could, the situation is different.

[8] If you see the knowledge gathering as part of the process, this sentence doesnt apply anymore. Btw, thats a nice model too.

[9] I enjoyed this metaphor first in a talk by Richard Rorty at Hamburg Literaturcafe some years ago.

[10] A more detailed approach by Eliezer Yudkowsky: The simple truth.
http://yudkowsky.net/rational/the-simple-truth